Quality by Design (QbD)
Myth or Reality?
Girish Malhotra PE, President EPCOT International
In the United States, the FDA’s initiative on nudging the pharmaceutical industry to invent, develop and commercialize products using technologies that will result in product quality by design (QbD) is a challenging task. It is also a noble task that will have major business process implications and ultimately high financial impact on healthcare costs.
Through presentations and pilot programs, the FDA is making its case to move the industry toward a win-win situation for all involved, especially consumers. These outlined items if submitted would assist in the approval process and allow continuous improvements. Since this is uncharted territory for drug developers and reviewers, it is necessary for the developers to present information that will convince the reviewers that the process will produce QbD “the desired state” rather than quality by inspection (QbI) “the present state”. Because this is a change in process and mindset, there will naturally be apprehension on the part of industry as it has traditionally worked in a defined comfort zone and is not sure about the value of change.
With that said, processes are developed and commercialized by chemists and engineers with the involvement of regulatory personnel to comply with the FDA, EPA and OSHA rules and regulations. In this evolution, one has to optimize the impact of their actions when it comes to the total business process. The current QbI methodology creates business processes and inventories that need infrastructure to support quality pharmaceutical production. This costs money and the consumers pay for it. QbD can be equated to Just in time (JIT) as the industry will have total control of the manufacturing process. Volumes have been written about the value of JIT and we all know that it smoothes out the total business process.
The FDA in its expectation of process analytical technology (PAT) framework and QbD stipulate a complete understanding of the interaction of raw materials and intermediates and control of process parameters. In a commercial operation, this will result in a QbD rather than QbI product. This is easier said than done. It requires combined application of knowledge, common sense and stepping out of bounds to establish a new paradigm.
In this paper, I am reviewing a chemistry outlined in a patent and sharing some of the opportunities to improve the manufacturing process. Process controls can be applied which will produce consistent quality product with no or minimal in-process testing. Similar methods and observations can be incorporated in the development of new chemistries so that we have a QbD process from the outset.
I have chosen production of an active ingredient rather than the formulation of the active pharmaceutical ingredient (API) with excipients. The reason for choosing reaction chemistry is its complexity and the length of time it takes to produce an API. In addition, much has been written about the formulation aspects and very little to none has been written about the manufacture of APIs. Thus, it needs proper consideration.
For the development of a process and chemistry of API’s an alternate approach could be considered. This is “out the box” thinking but worth consideration. Instead of having a mindset that we are developing a “pharmaceutical”, it might be easier to consider that we are developing a “specialty chemical” that might have pharmaceutical value. This should simplify many of the drug development processes.
In this alternate approach, once an optimum process has been developed for a specialty chemical where we know the interaction of each raw material and intermediates, the critical parameters and how to control them, every regulatory requirement can be included to meet the necessary standards. Since we have to apply regulatory requirements only on one process, the product and process development is simplified. Such an approach might also be a way to reduce the “time to market”. I believe that if we are able meet quality and performance specifications all the time, we might have to contend with fewer regulations also.
By using the “right” process we will produce a “quality” product, which in turn will reduce waste, work in process inventories, regulatory oversight and bureaucracy i.e. simplify the business process. Doing it right the first time, by a repeatable process is the key and has to be the method of choice. If we are able to accomplish this from the outset, we would not have to live with 2.5 Sigma processes, and the process of continuous improvement would be less expensive compared to after the fact improvements.
The following considerations are necessary in the development, simplification and commercialization of the “right” process. Most of these are being used in the manufacture of chemicals. The understanding and application of these also allows the control of processes using commercially available process control technologies. The following are taught in chemical engineering curriculum, thus these are not new.
1. Total process feasibility. Each unit process step has to be reviewed individually and collectively.
2. Is the stoichiometry optimized?
3. Are the heat and mass balance optimized?
4. Are the reaction kinetics understood and applied to simplify the process?
5. Are proper unit operations being used?
6. Are the necessary steps in place to reduce the cycle time?
7. Can a single solvent be used for the whole process? This economizes solvent recovery and the related investment.
8. Can we eliminate isolation of intermediates?
9. Are the raw materials to be used easy to handle?
10. How can the phase separation be improved and simplified, if it is part of the process?
11. How can I improve the conversion of each process step? Lower conversion means that there is raw material loss, which has to be recovered and/or treated in the effluent system or disposed of as hazardous waste. Lower conversion also means that unless the unconverted raw materials or impurities are removed prior to the subsequent reaction steps, additional impurities will be created adding to the process complexity.
12. Are the safety requirements met and is the process safe?
13. If the developers were operating the process, what process modifications and/or additions would be included to have the simplest process?
14. Is the process meeting all of the environmental standards?
15. Is the rework eliminated and/or minimized?
16. Is the process economical? A thorough understanding of every interaction allows one to have a complete grasp of the impact of every process change and its influence on the product quality. If above considerations are followed all the time QbD becomes a natural part of the development process. In addition to the above considerations, each chemistry and process has its nuances and if recognized and implemented can simplify the manufacturing processes further. The incorporation also allows one to have complete control of the process. One can react to any unexpected changes and deliver quality. From my experiences, it also allows one to repeat the mistakes. If this can be done, developers will have done an excellent job. It would be like driving a car smoothly under every condition. Today’s pharmaceutical manufacturing can be compared to an automobile driving us vs. us driving the automobile.
The application of the above considerations is part of the reviewed process. It is also possible that some of the process observations mentioned below if implemented can lead to a continuous process. It is well known that a continuous process is more economical than a batch process and produces consistent quality i.e. QbD.
Dr. Moheb Nasr, Director, CDER’s Office of New Drug Quality Assessment, Dr, Janet Woodcock, Dr. Scott Gottlieb and others at the FDA have echoed the sentiment well know in the chemical industry that “the pharmaceutical industry can only realize the full benefit of QbD by developing and implementing continuous processing”.
US patent 4,623,736
This patent covers the synthesis of ibuprofen and naproxen. Patent describes different routes. However, I have only reviewed the preparation of ibuprofen from example 1. Similar observations can be applied for the preparation of naproxen (example 2) also.
Process description: Isobutyl benzene is reacted with alpha-chloropropionyl chloride and aluminum chloride with methylene chloride as a solvent. The reaction is carried out at 0 to – 5°C. Excess aluminum chloride is neutralized with dilute hydrochloric acid. Organic and aqueous phases are separated. The organics in the water phase are extracted with methylene chloride. The combined organic phase is washed with sodium bicarbonate to pH 7-8. Organic phase is heated to remove methylene chloride and heptane is added. The resulting ketone is then reacted with neopentyl glycol in presence of concentrated sulfuric acid. The material is refluxed at about 97-107°C. Water is azeotroped. After the reaction is complete, about 8 hours, the reaction mass is washed with dilute solution of sodium bicarbonate. All of the organic phases are combined and washed with water. Water and organic phases are separated. Heptane is removed under vacuum and ketal oil is obtained.
Ketal oil is heated to about 140°C and this removes any remaining heptane. Catalytic amount of Zinc 2-ethylhexanoate dissolved in heptane is slowly added and the temperature is maintained at 140-150°C. Since the reaction temperature is above the boiling point of heptane, all of the heptane boils off. Upon reaction completion, the liquid is cooled to about 25°C and a filter-aid is added to absorb the zinc salt. Zinc salt is filtered. Liquid is carbon treated and filtered. Heptane used for solid washing is combined with the process liquid.
Chloroester produced above is heated to 95-100°C and sodium hydroxide solution added. The reaction mass was heated to about 95°C. Once the hydrolysis is complete, water is added to dissolve the sodium salt in water. Aqueous solution is cooled to 0°C and crystals of sodium ibuprofen are filtered, washed with heptane and dried.
Sodium salt is converted to the acid by dissolving it in water and addition of heptane and hydrochloric acid. Some of the heptane is distilled and the acid is crystallized from heptane.
Review: The above synthesis is classical chemistry and if one did not know that the final product 2-[4-(2-methylpropyl) phenyl] propanoic acid has a pharmaceutical value; it would be treated as a specialty chemical synthesis. Since it is a pharmaceutical, it needs to meet all of the requirements for human consumption. We have to keep in mind that most of the patents are based on laboratory procedures and for commercial production; we need to have a viable and economic process. Thus, from a business perspective the process might need modifications to be an economic commercial operation. Considerations mentioned above should be applied to each step individually and collectively to the whole process.
Step 1: In the first step of the reaction, there are few opportunities. The Friedel-Crafts reaction is carried out with methylene chloride as the solvent. After neutralization, a new solvent (heptane) is added and it is used in the rest of the process. If we can substitute heptane for methylene chloride from the beginning, only one solvent will be used. This reduces investment necessary for methylene chloride tank, recovery and handling. Use of heptane can lead to a price advantage due to purchase of larger quantity of single solvent.
Solid handling of aluminum chloride requires special methods. Aluminum chloride can be slurried in heptane, fed as a liquid and this can simplify processing and handling. Reaction is carried out at 0 to -5°C. Most of the Friedel Crafts reactions are zero order reactions. To control the exotherm, the process as described suggests slow addition of propionyl chloride. In addition, proper feeding system with feed, temperature and other controllers would be needed to control the kinetics and temperature. It is well known that if a reaction can be conducted at higher temperatures, the reaction time can be reduced. With a properly designed system, it is possible to have a continuous process for this step. [This could be labeled unsafe but since I have practiced it I can write about it]. Commercially available process controllers are used on a routine basis. A properly designed reaction control system will have high conversion and might not require in-process testing of conversion. In a continuous process, phases can be separated continuously without any interface controls. All these improve productivity.
The process as suggested uses about 9% excess of alpha-chloropropionyl chloride and 28% excess of aluminum chloride. If the raw material use can be optimized and it will reduce chemicals needed for neutralization. This will also reduce raw material and waste treatment costs and improve productivity.
Step 2:In the second step, product from step 1 is reacted with Neopentyl glycol to produce a ketal. The mixture is heated to about 90°C and the condensation reaction is carried in the presence of concentrated sulfuric acid. Mixture is heated to 97-107°C and the produced water is azeotroped. The reaction is complete in about 8 hours. The mixture is cooled and neutralized with sodium bicarbonate solution. Each phase is washed and heptane is removed from the organic phase under vacuum. Glycol could be added as a molten liquid. With proper heat and mass balance, the condensation can be accelerated and controlled, acid neutralized and heptane removed. It is possible to do the condensation continuously with proper control of glycol addition. The process as suggested uses about 39% excess glycol based on isobutyl benzene. There is an opportunity to reduce this amount and therefore lower cost.
Step 3:Ketal from step 2 is converted to ibuprofen ester using zinc 2-ethylhexanoate. The reaction is carried out at about 135-150°C. Rearrangement reaction is complete in about two hours. The reaction mass is cooled to 20-30°C and a filter aid is added. Heptane is added and the solids filtered out. Filtered liquid is carbon treated and filtered. Since ketal and the catalyst are liquids, processing and controls can be done by classical methods. It is possible that the reaction time can be reduced significantly to have a continuous process that would include carbon treatment and filtration.
Steps 4 and 5: These are classical cases of producing a sodium salt through crystallization and converting the sodium salt to its acid, crystallization, filtration and drying. Steps 4 & 5 are routine unit processes and operations that can be done continuously with proper process controls to produce a product that meets quality specifications all the time.
Overall yield, %
|Yield per step, %||95||80||70|
|Number of Steps|
Overall comments: Once the steps of a continuous ibuprofen process are assembled, it might be possible to achieve additional improvements i.e. yield improvement due to changes in addition methods and operating conditions. This is all possible, as the developers have done a good job of keeping the chemistry and processing of ibuprofen and naproxen simple. Again, in order to improve productivity, cost and achieve quality, developers, chemists and engineers have to review each reaction step individually and collectively and simplify them to have the most productive process. Mr. Thomas E. Burakowski from Boehringer-Ingelheim Pharmaceuticals, Inc. in a recent article mentions that in the manufacture of pharmaceuticals it is not uncommon to have 20-30 synthesis steps. If the yield in each of the 20 steps is 98% (highly unlikely), the overall process yield would be about 66.8 %. I believe processes with this many steps would have a long cycle time and would be uneconomical in the specialty chemical world. Since, we need to produce the API, we have opportunities. Table 1 shows the overall yield of processes that have 5 or 10 steps with 95, 80 or 70% yield for each reaction step. Yields are based on the key starting material. Lower yields suggest cost reduction, productivity improvement and quality enhancement opportunities. Every effort should be made to minimize the yield loss. Higher yield has to be the goal of every process. In the pharma world, we are able to recover such losses though price of the drug as people will pay to extend their life. However, in the pure chemical world such processes would be considered uneconomical.
We have acknowledged “quality by inspection” as an acceptable business model for life-prolonging drugs and companies have been able to achieve their profit margins. Thus, there is very little incentive to optimize drug manufacturing processes. The FDA has echoed this in its assessment of the current manufacturing technology1. Lack of optimization results in variable product quality. We are able to meet stringent quality standards through isolation, purification and checks at every step. This is very similar to steps that are taken in a laboratory or pilot plant to develop theses products. One could call pharma manufacturing a large-scale laboratory or pilot plant. Since we have to inspect quality, the whole business process is burdened with unproductive costs.
There are opportunities in every synthesis and one has to capitalize on them from the outset. It is well recognized that there is time pressure for first to market but lack of an optimum process burdens the business process and no one has ever calculated the economic impact of this burden. Lowering or removal of these burdens will lower costs and improve profits further.
A brief review of several patents suggests simplification opportunities2. Review of Ganciclovir intermediate manufacture3 suggests that the synthesis can be modified and simplified to produce only the desired isomer only. This can create a new process, which could out the bounds of these patents, and QbD product. Manufacture of Modafinil (Provigil)4 for the treatment of narcolepsy can be simplified to a continuous process. Similarly Omeprazole ® (Prilosec) could be produced using a continuous process.
A complete knowledge of every interaction of raw materials and process conditions will produce quality with very few or no isolation of intermediates and in-process tests, a real QbD. Thus, QbD is not a myth. It is feasible and achievable.
Drive to achieve QbD for any API being developed in the laboratory should not start or end at the understanding of the interaction of raw materials, their reaction conditions and physical properties for chemicals that are used to produce any product. It can be also be applied to the existing API’s also as it will simplify the total business process and improve profitability.
It may be time to explore alternate ways to simplify process development and manufacturing methods for APIs so that after drug efficacy, “first time quality” becomes a driving force in pharma manufacturing. Any company achieving QbD will have the optimum process (i.e. lowest cost) for the used chemistry. This can act as a deterrent to the entry of generic producers. In addition, it will keep the profitability at a significantly high level beyond the patent life compared to the current scenario.
1Innovation and Continuous Improvement in Pharmaceutical Manufacturing. The PAT Team and Manufacturing Science Working Group Report. http://www.fda.gov/cder/gmp/gmp2004/manufSciWP.pdf